1. Technical Field
The present invention relates to a resource management system, a resource information providing method and a program. More particularly, the present invention relates to a resource management system, a resource information providing method and a program for providing resource information in a distributed information system.
2. Background Art
Conventional resource information management systems for clusters can execute processes for collecting information from all the nodes stably at a high speed because the nodes have a uniform performance and are connected to one another by way of a high speed network. However, in the case of a system for consolidating and managing a plurality of clusters connected by way of a wide area network or a plurality of personal computers of different types, it is time consuming to acquire resource information particularly when the scope of retrieval of resource information is large and the processing time may vary depending on the object of retrieval.
An MDS (monitoring discovery system) described in Non-Patent Publication 1 is known as resource information management system in a wide area distributed system. An MDS manages resource information by means of a directory and, when a retrieval process takes place for resource information, it acquires the required resource information and caches the same in the directory. When another retrieval process takes place for the same resource information, the cached resource information can be provided to the new retrieving party to reduce the retrieving time. The effective period of the cache is defined in advance for the resource information collected by the MDS. When a client requests for information after the expiration of the effective period, the cached information is destroyed and a process for acquiring the most recently updated information is executed. A plurality of MDSs can be linked to one another in a hierarchical manner. Thus, a variety of different systems can be realized by using MDSs to cover a single cluster and large scale distributed systems.
Patent Publication 1 describes a technique wherein a data updating method uses a network server. With the described technique, the frequency of access to the data held in a cache by client terminals is detected and the data is read out from a main server before the holding period of the cache expires to update the data held in the cache. With this arrangement, it is possible to automatically update the cache data that can be expected to be accessed by client terminals in the future, whereby the most recently updated data can be transferred to a client terminal in a short time.
Patent Publication 1: JP-2001-175527-A (FIG. 1, Summary)
Non-Patent Document 1: Grid Information Services for Distributed resource
Sharing, Czajkowski, K.; Fitzgerald, S.; Foster, I.; Kesselman, C.; High Performance Distributed Computing, 2001. Proceedings, 10th IEEE International Symposium on, 7-9 Aug. 2001